524 
ASTRONOMY: F. H. SEARES 
be given to the following points : First, if there are differences of grad- 
ation between the blue and yellow images, the exposure-ratios of bright 
and faint stars of the same color will be different. In other words, the 
ratios will depend upon the size of the images. The matter is easily 
investigated by observing repeatedly the same stars with different 
apertures and exposure times; the variation of exposure-ratio with the 
size of image can thus be determined. In practice it will be advantage- 
ous, wherever possible, to adjust exposure and aperture so that the 
resulting yellow image is of a standard size. Preliminary tests indicate 
that the systematic gradation difference for the blue and yellow images 
is small. Small deviations from the standard image are therefore of 
no consequence, and larger differences are readily taken into account 
in the reductions. 
Second, atmospheric extinction reduces the intensity of blue light 
more than that of yellow and therefore modifies the exposure-ratio. 
Although the effect enters differentially, its influence is appreciable, 
except for small zenith distances. Tables of corrections can be derived 
by determining the exposure-ratio for the same star at different zenith 
distances. 
Third, care should be taken to bring the plates into equilibrium with 
the atmospheric conditions before beginning the exposures; otherwise 
the absorption of moisture during the exposure may introduce vari- 
ations of sensitiveness and gradation which are capable of influencing 
seriously the results. 
No extensive test of the method has yet been made, but it has been 
tried sufficiently to demonstrate its practicability. About 40 plates, 
mainly of individual bright stars, were exposed on three successive 
nights, two or more plates being obtained on each of 10 different stars. 
Two yellow and five blue exposures were made on each plate; further 
three exposures through a filter transmitting to the violet of X 4000 
were also added in each case in order to test the precision of results 
derived from different regions of the spectrum. To avoid gradation 
errors, the apertures and exposure times were adjusted so that the 
images dift'er little from a standard size. 
For the plate and filter used the exposure ratio of blue to yellow varies 
from about i to J for spectral types A to M. That of blue to ultra- 
violet would be a much smaller quantity but for the fact that the large 
exposure factor of the ultra-violet filter was offset by a change in the 
aperture — ^the exposures for blue and yellow always being made through 
a wire-gauze screen absorbing about 3 magnitudes. 
